CPM calculator

Ad spend is the media cost for the period, in a single currency. Nothing here converts, and an average across two currencies without a stated rate is a figure nobody can check.

Impressions is the number of times the ad was served, as the platform counts it. Platforms do not all count it the same way — served and viewable are different events, and a placement that renders below the fold may be counted before anybody could have seen it. Two platforms reporting the same cost per thousand may be describing two different things, and the gap is not visible in the number.

Clicks is optional. Supply it and the calculator produces the cost per click and the click-through rate, which is what turns a price into a reading.

The cheapest thousand impressions in any auction are almost always the ones the rest of the market has decided are worth least: broad placements, low-attention surfaces, audiences with no particular reason to care. Buying them lowers this number immediately. It rarely lowers the cost of an outcome.

The arithmetic is worth doing once, because it settles the argument. Cost per click is the cost per thousand divided by ten times the click-through rate. Halve the price of the impression and halve the rate at which people click, and the cost per click is exactly where it started, with twice as many impressions burnt to get there. The headline metric improved, the creative reached more people who ignored it, and nothing that matters moved.

That relationship is also why the price is a poor creative test. A campaign whose cost per thousand rose while its cost per click fell is usually working better, not worse: the auction charged more for attention that converted at a higher rate. An account managed on the impression price alone will systematically prefer the first campaign to the second.

The auction, mostly. The price is set against whoever else wants the same people at the same moment. Retail peaks, election periods, a competitor's launch and the last week of a quarter all raise it without a single change on your side.

Audience definition. A narrow audience of high-value people is contested by everyone who values them, and it costs accordingly. Widening a definition usually lowers the price and changes who is reached, which is a trade rather than a saving.

Placement mix. Premium, high-attention surfaces cost more per thousand than the long tail of the same platform's network, and a campaign left on automatic placements will drift toward the cheap end of that mix on its own.

Format and duration. Longer video, larger canvases and formats that demand more attention are priced as such.

Quality signals. Most platforms discount inventory for ads that people engage with and surcharge ads they hide or report. A creative problem can therefore appear first as a price problem, which is one of the few cases where this metric is genuinely diagnostic.

The useful habit is to treat cost per thousand as the denominator of a story rather than the story.

Price up, rate up, cost per click down. The buy got more expensive and better. This is usually the shape of a creative that started working, and it is the case most often misread as a regression.

Price down, rate down, cost per click flat. More impressions, same result. Nothing was gained and reach was spent.

Price flat, rate falling. Creative fatigue, or an audience that has been reached too often. The price cannot tell you which; frequency and the age of the creative can.

Price rising sharply with no change of your own. Look outward before looking inward. Seasonality and competition move this figure more than anything inside the account, and rebuilding a working campaign in response to the market's own price is a common and expensive mistake.

Sometimes it genuinely is the price. When the click-through rate has held and the cost per thousand has risen enough to break the maths, there are three honest moves.

Change when you buy. Auction pressure is not flat across the week or the year, and a campaign with no deadline can often be bought at a materially lower price simply by avoiding the period everybody else is bidding into.

Change where you buy. Excluding the placements that consistently produce impressions and nothing else raises the average price of what remains and lowers the cost of the outcome, which is the trade that matters.

Change the definition of the audience. If a narrow segment has become contested to the point where the arithmetic no longer works, a wider one that still contains the people you want is a cheaper route to them than outbidding the market for the small version.

What does not help is optimising toward the cheapest thousand impressions available. That target is always reachable, and reaching it costs exactly as much as the outcomes it fails to produce.